Olfactory enrichment mitigates delayed neurocognitive recovery after major orthopaedic surgery in older patients: A randomised controlled trial

ABSTRACT

Background

Delayed neurocognitive recovery (dNCR) is a prevalent complication in older patients undergoing surgery. It may progress to long‐term cognitive impairment and increase the risk of Alzheimer’s disease.

Aims

This study aimed to evaluate the effects of olfactory enrichment on dNCR and to examine the association between olfactory function and dNCR.

Methods

This sham‐controlled, assessor‐blind, parallel‐group randomised trial enrolled 149 participants aged 65 or older undergoing elective total knee or hip replacement under general anaesthesia. Participants were assigned to either the olfactory enrichment group or the sham group. The intervention group received daily olfactory enrichment from 3 days preoperatively to 7 days postoperatively. Cognitive function was evaluated using a neuropsychological test battery 3 days before and 7 days after surgery. Olfactory identification ability was assessed by five‐odour olfactory detection arrays. Propensity score matching analysis was employed to mitigate potential confounding and selection bias.

Results

A total of 131 patients completed the study (62 in the olfactory enrichment group and 69 in the sham group). The overall incidence of dNCR was 26.7% (35 out of 131). In the intention‐to‐treat analysis, the difference between groups was not statistically significant (19.4% vs. 33.3%; χ ² = 3.259; p = 0.071). However, in the 1:1 propensity score–matched cohort (n = 82), the incidence of dNCR was significantly lower in the olfactory enrichment group than in the sham group (12.2% vs. 39.0%; χ ² = 7.476; p = 0.005). Raw postoperative cognitive scores and individual change scores did not differ between the groups. Participants with decreased olfactory identification scores (n = 32) had a significantly higher incidence of dNCR than those with stable or improved scores (40.6% vs. 22.2%; χ ² = 4.183; p = 0.041).

Conclusions

In older patients undergoing major orthopaedic surgery, perioperative olfactory dysfunction is associated with an increased risk of dNCR. Olfactory enrichment may represent a potential nonpharmacological strategy for reducing postoperative cognitive decline in this population.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Postoperative neurocognitive disorders are among the most common complications in older adults undergoing surgery and include postoperative delirium, delayed neurocognitive recovery (dNCR; cognitive decline within 30 days after surgery) and postoperative neurocognitive disorder (cognitive decline from 30 days to 12 months after surgery). These conditions are associated with long‐term cognitive impairment, an increased risk of Alzheimer’s disease and related dementia, as well as premature mortality. However, effective preventative and therapeutic interventions remain limited.

WHAT THIS STUDY ADDS

• In older patients undergoing major orthopaedic surgery, perioperative olfactory dysfunction is associated with an increased risk of dNCR. Perioperative olfactory enrichment may effectively reduce the incidence of dNCR in this population.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings support further clinical trials to evaluate the efficacy of olfactory enrichment as a nonpharmacological strategy for preventing postoperative neurocognitive disorder.

INTRODUCTION

Postoperative neurocognitive disorders are among the most common complications in older adults undergoing surgery and include postoperative delirium, delayed neurocognitive recovery (dNCR; cognitive decline within 30 days after surgery) and postoperative neurocognitive disorder (cognitive decline from 30 days to 12 months after surgery). ¹ , ² Previous studies reported that the incidence of dNCR ranged from 18% to 40%. Although initially presenting as short‐term cognitive impairment, dNCR may progress to long‐term cognitive deficits and is associated with an increased risk of Alzheimer’s disease (AD) and related dementia, as well as premature mortality. ³ , ⁴ , ⁵ These outcomes can substantially impair quality of life and may increase the risk of physical and psychological comorbidities, ⁶ , ⁷ , ⁸ underscoring the importance of early detection and timely intervention. However, effective preventative and therapeutic interventions for dNCR remain limited.

Olfactory dysfunction is increasingly recognised as an early feature of several neurodegenerative diseases. Impaired olfaction has been reported in Parkinson’s disease and AD and is often associated with poorer cognitive performance. ⁹ , ¹⁰ , ¹¹ , ¹² Preserved olfactory function is generally associated with better cognitive performance, whereas olfactory deficits often appear as early manifestations of neurodegeneration. ¹³ Specifically, olfactory identification, discrimination and detection thresholds are associated with episodic and semantic memory ¹⁴ as well as executive function. ¹⁵ Moreover, accumulating evidence also suggests that olfactory stimulation can positively impact cognitive function, ¹⁶ indicating that olfactory function may represent a potential target for interventions aimed at improving cognitive outcomes.

Olfactory enrichment, also referred to as olfactory training, is a safe and cost‐effective intervention involving systematic exposure to a structured set of odorants. ¹⁷ , ¹⁸ Olfactory enrichment can improve olfactory dysfunction of various aetiologies ¹⁹ and may attenuate cognitive impairment. ²⁰ Olfactory enrichment is also associated with increases in the volume of olfactory‐related brain regions, including the olfactory bulb and hippocampus, and altered functional connectivity. ²⁰ Animal studies have suggested that surgery and anaesthesia may induce olfactory impairment that contributes to dNCR, whereas olfactory enrichment can reduce postoperative cognitive impairment. ²¹ However, whether olfactory enrichment can mitigate dNCR in humans remains unclear. We, therefore, conducted a randomised clinical study in older adults undergoing major orthopaedic surgery to evaluate the effects of olfactory enrichment on dNCR and to examine the association between olfactory function and dNCR.

METHODS

Study design

This sham‐controlled, assessor‐blind, parallel‐group randomised trial was conducted in Shanghai Tenth People's Hospital. Written informed consent was obtained from all participants. This trial was registered at ClinicalTrials.gov (NCT03441074).

Participant screening and enrolment

Participants were screened on hospital admission. Preoperative assessments were conducted using the methods outlined in our previous study, ²² , ²³ including review of medical records and direct patient interviews. Individuals who completed these preoperative assessments and met the eligibility criteria were invited to participate.

Patients were eligible for inclusion if they were aged ≥ 65 years, were expected to have a postoperative hospital stay of at least 7 days, had an American Society of Anesthesiologists physical status classification of I–II, were native Mandarin speakers and were scheduled to undergo total knee or hip replacement under general anaesthesia. Patients were excluded if they had a prior diagnosis of neurological disease according to the International Statistical Classification of Diseases and Related Health Problems, 10th Revision; a history of mental disorders diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders; a history of nasal or sinus disease or surgery; a common cold within the previous week; visual or auditory impairment that could interfere with study assessments; unwillingness to comply with the study protocol or procedures; or cognitive impairment based on the Mini‐Mental State Examination (MMSE, score < 18 for illiterate participants, < 20 for primary school education or < 24 for secondary school education or higher). ²⁴

Preoperative assessments

Preoperative assessments were performed on the day of hospital admission, usually 3 days before surgery. The assessments included demographic characteristics (e.g., age, sex, height, weight and years of education), medical history and comorbidities. Preoperative cognitive function was assessed using a neuropsychological test battery, and olfactory function was assessed using an odour identification test.

Randomisation

Randomisation and allocation were overseen by a researcher who was not otherwise involved in the trial. Eligible patients were randomised in a 1:1 ratio using random‐sized blocks. Individual allocations were accessed by investigators through a secure website after consent was obtained and no earlier than the morning of surgery to conceal allocation as long as practical. The randomised treatments were administered by clinicians in coordination with research personnel. All outcome assessments were conducted by separate research assistants who were blinded to treatment assignment.

Participants were informed that they would receive various inhaled substances. Odourless tea bags (sham) and scented tea bags (intervention) were identical in appearance and prepared by an independent researcher who maintained the randomisation list. Outcome assessors and data analysts were blinded to group allocation; they had no access to the randomisation file and interacted with participants in a neutral‐smelling room. Treatment could be discontinued at the request of participants or clinicians; however, unblinding was permitted only in medical emergencies (e.g., suspected allergic reaction to an inhaled scent), in which case, the treating clinician opened a sealed opaque envelope and reported the event to the data and safety monitoring board.

Postoperative interview

Seven days after surgery, participants were reassessed using the same neuropsychological test battery and odour identification test as in the preoperative interview. Complications and mortality were recorded.

Normalisation control group

In studies that use repeated neuropsychological tests, practice effects and natural variation in cognitive test performance can result in misinterpretation of the outcome. To account for these effects, 30 age‐ and sex‐matched volunteers from Shanghai, China, were recruited through a community newspaper. The volunteers completed the same neuropsychological tests, administered by the same investigators at the same time intervals as the trial participants but did not undergo surgery or anaesthesia.

Outcome measures

Primary outcome

Incidence of dNCR

Participants underwent a neuropsychological test battery adapted from the International Study Group of Postoperative Cognitive Dysfunction battery and modified to reflect the characteristics of Chinese patients. The tests were administered in a quiet place in the general wards 3 days before and then 7 days after surgery. The battery was designed to measure memory, psychomotor speed and dexterity, physical motor speed, attention capacity and perceptual‐spatial function. ²⁵ It consisted of the Hopkins Verbal Learning Test–Revised (HVLT‐R), Brief Visuospatial Memory Test–Revised (BVMT‐R), Trail Making Test, Digit Span Test, HVLT‐R delayed recall test, HVLT‐R Recognition Discrimination Index, BVMT‐R delayed recall test, BVMT‐R Recognition Discrimination Index and verbal fluency test. To minimise practice effects, two parallel versions of the neuropsychological battery were used (A preoperatively, B postoperatively), with minor variations between versions (e.g., different HVLT‐R word lists). ²⁶

dNCR was defined according to the International Study of Postoperative Cognitive Dysfunction ⁶ using a Z‐score method. For each test, the Z value was calculated as (ΔX − ΔX _C)/SD_ΔXC, where ΔX is the individual change score (postoperative minus preoperative), ΔXC is the mean change in healthy controls (representing systematic error), and SD_ΔXC is the standard deviation of that change in controls (representing expected variability). A participant was classified as having dNCR if the absolute Z‐score was ≥ 1.96 on at least two different tests. ⁶

Secondary outcome

Odour identification ability

Odour identification ability was assessed using a five‐odour identification array ²⁷ administered 3 days before and approximately 7 days after surgery. This test consisted of five unlabelled pen‐shaped test sticks, each containing five odorants, including vinegar, banana, mint, rose and coal tar odour. ²⁸ Participants were asked to freely identify each odour after sniffing, with an interval of approximately 30 s between odour presentations. Scores ranged from 0 to 5.

Study interventions

Olfactory enrichment

Participants randomised to the olfactory enrichment group were exposed to aromatic fragrances daily from 3 days before to 7 days after the surgery. Four odorants were used in rotation: phenylethyl alcohol (rose), eucalyptol (eucalyptus), citronellal (lemon) and eugenol (cloves). ¹⁸ Each odorant was changed after 24 h of exposure. Ten drops of 100% essential oil (0.5 mL) were applied to a fabric insert in tea bags hung at the patient’s bedside, providing passive exposure throughout the day. ²⁹ , ³⁰ , ³¹ Participants were informed that the study involved exposure to various inhaled substances, some of which might have familiar pleasant odours, thereby maintaining sufficient ambiguity regarding group allocation.

Sham treatment

Patients assigned to sham treatment followed the same schedule and session duration as those in the olfactory enrichment group. However, the tea bags provided to the sham group were odourless. The tea bags appeared identical to those used in the olfactory enrichment group.

Anaesthesia and surgery

Surgeries included total knee replacement and total hip replacement. Anaesthesia care was conducted in accordance with the American Society of Anesthesiologists guidelines, hospital policy and at the discretion of anaesthesiologists. Anaesthesia protocols were standardised, and no anticholinergic agents were used as premedication. On arrival, all patients received standard monitoring. Anaesthesia was induced with midazolam (1–2 mg) for anxiolysis; propofol (2 mg/kg), sufentanil (0.5–1.0 μg/kg) and cisatracurium (0.5 mg/kg) for amnesia, analgesia and muscle relaxation. Atropine (0.25–1.00 mg) was administered to all patients to reduce airway secretions. Anaesthesia was maintained with propofol (4–12 mg/kg/h) alone or in combination with sevoflurane (1%). Anaesthetic depth was monitored using the bispectral index. ³²

Harm definition and assessment

During the study, potential harms were monitored systematically. Harms were defined as any adverse events or negative outcomes related to the olfactory enrichment or the study procedures. At each postoperative visit, participants were asked about any new or worsening symptoms since the previous assessment, including but not limited to respiratory issues, gastrointestinal disturbances, allergic reactions and psychological symptoms. Medical records were also reviewed by the clinical team for any unsolicited reports of adverse events, including emergency interventions or additional treatments that might indicate a study‐related adverse event.

In addition, an open line of communication was maintained with participants and their caregivers throughout the study period. Participants were encouraged to report any concerns or discomfort they experienced at any time, including outside scheduled visits. This allowed the capture of unexpected or infrequent adverse events that might not have been captured during routine assessments.

Any reported adverse events were thoroughly investigated by the study investigators. The severity of each event and its potential association with the study intervention were assessed. If an event was considered related to the study, appropriate medical care was provided, and the event was documented in detail, including the nature of the harm, time of onset, duration and actions taken.

Statistical analysis

Sample size

The incidence of dNCR within 7 days after surgery in patients undergoing elective hip or knee surgery is estimated at 20%–40%. ⁵ , ³³ Assuming a two‐sided χ ² test (α = 0.05, power = 0.80) and expecting olfactory enrichment to halve the dNCR incidence from 30% to 15%, ³⁴ 47 participants per arm were required. Allowing for a 10% drop‐out, 104 participants were targeted for randomisation.

Outcome analysis

The statistical analysis plan was finalised prior to data analysis. The Kolmogorov–Smirnov test was used to assess normality. Homoscedasticity was evaluated using residual plots and Levene’s test, which confirmed equal variances across groups (p > 0.05). Normally distributed continuous variables are presented as mean (standard deviation [SD]) and non‐normally distributed variables as median (interquartile range [IQR]). Differences between patients who did and did not develop dNCR were assessed using Student's t‐test or Mann–Whitney U test, as appropriate. Categorical variables are presented as frequencies and proportions and assessed using chi‐square or Fisher's exact test.

Propensity score matching analysis

To minimise the effects of potential confounding factors, a propensity score matching (PSM) analysis was applied. ³⁵ Propensity scores were calculated for each patient using the predicted probabilities from the multivariable logistic regression model, regardless of the statistical significance of the independent variables in the model. Patients were matched 1:1 without replacement using a nearest‐neighbour approach with calliper restrictions. PSM was performed using SPSS version 26.0 (SPSS Inc.).

Multiple regression analysis

To further mitigate the impact of confounding bias in the matched cohort, multivariable logistic regression analyses were conducted to assess the effect of olfactory enrichment on dNCR. Logistic regression models were adjusted for age, sex, years of education and baseline MMSE scores. ³⁶

SPSS version 26.0 (SPSS Inc.) was used to analyse the data. A two‐sided p < 0.05 was considered statistically significant for all analyses.

RESULTS

Participant characteristics

A total of 435 patients aged 65 years or older were screened from February to December 2018. Of these, 286 were excluded because they did not meet the inclusion criteria (n = 216) or declined the required preoperative cognitive assessment (n = 70). The remaining 149 participants were enrolled and randomised into the olfactory enrichment group or the sham group. After consent, nine participants were excluded due to a change in anaesthesia plan (n = 4) or cancellation of surgery (n = 5), and a further nine declined postoperative assessments. There were no missing data for variables of interest, and no unintended adverse events were reported in either group (figure 1).

FIGURE 1.

The flowchart of the study.

In the full intention‐to‐treat cohort, the median (IQR) age was 71.0 (68.0–75.0) years in the enrichment group and 73.0 (68.0–77.0) years in the sham group. Both groups were predominantly female (66.1% vs. 78.3%). The mean (SD) body mass index (BMI) was 26.5 (3.7) kg/m² in the enrichment group and 25.8 (3.9) kg/m² in the sham group, and the median educational attainment was 9 years in both groups. Baseline global cognitive scores differed between groups (MMSE 25.0 [23.0, 27.0] vs. 27.0 [24.5, 29.0]; Z = −3.050; p = 0.002), although all participants exceeded the study entry thresholds, indicating generally intact preoperative cognition. Similar patterns were observed across domain‐specific tests (BVMT‐R, verbal fluency test, etc.), with slightly lower scores in the enrichment arm. Because these baseline differences could influence the interpretation of cognitive outcomes, PSM was performed to balance the groups. After PSM, 41 matched pairs were obtained. The matched pairs were comparable in age, sex distribution, BMI, educational attainment and all baseline cognitive measures (all p > 0.05), providing a well‐balanced cohort for the primary outcome analysis (table 1).

TABLE 1.

Baseline characteristics of study participants

	All eligible patients				Propensity score matched
	Intervention (n = 62)	Sham (n = 69)	Statistic	p value	Intervention (n = 41)	Sham (n = 41)	Statistic	p value
Age (years), median (IQR)	71.0 (68.0, 75.0)	73.0 (68.0, 77.0)	Z = −0.640	0.522	72.0 (69.0, 74.5)	72.0 (68.0, 77.0)	Z = −0.279	0.845
Sex, male, n (%)	21.0 (33.9)	15.0 (21.7)	χ 2 = 2.412	0.120	14.0 (34.1)	7.0 (17.1)	χ 2 = 2.381	0.077
BMI (kg/m2), mean (SD)	26.5 (3.7)	25.8 (3.9)	t = 0.858	0.340	25.6 (3.4)	26.1 (4.1)	t = −0.301	0.598
Education (years), median (IQR)	9.0 (6.0, 12.0)	9.0 (6.5, 9)	Z = −0.038	0.969	9.0 (7, 12)	9.0 (6.0, 9.0)	Z = −1.207	0.151
Baseline MMSE (scores), median (IQR)	25.0 (23.0, 27.0)	27.0 (24.5, 29.0)	Z = −3.050	0.002	26.0 (23.5, 27)	26.0 (23.5, 27.0)	Z = −0.229	0.758
ADL (scores), median (IQR)	15.0 (14.0, 17.0)	14.0 (14.0, 16.0)	Z = −1.927	0.054	15.0 (14.0, 17.5)	14.0 (14.0, 16.0)	Z = −1.515	0.130
Propofol (mg), median (IQR)	711.4 (536.3, 962.6)	816.0 (533.3, 1047.0)	Z = −0.470	0.638	810.0 (595.5, 1170.0)	729.0 (540.0, 924.8)	Z = −1.429	0.153
Sufentanil (μg), median (IQR)	51.1 (41.3, 64.3)	49.5 (35.2, 60.0)	Z = −1.406	0.160	48.15 (41.9, 58.5)	49.6 (34.5, 60.4)	Z = −0.176	0.860
Cisatracurium (mg), median (IQR)	35 (31.5, 37.5)	32.5 (29.5, 37.5)	Z = −1.632	0.103	32.5 (29.8, 36.1)	32.1 (28.5, 32.5)	Z = −0.721	0.471
Preoperative cognitive function
HVLT‐R (scores), median (IQR)	11.0 (9.0, 12.3)	12.0 (9.0, 15.0)	Z = −1.742	0.085	11.0 (9.0, 13.5)	12.0 (8.5, 15.0)	Z = −0.372	0.710
BVMT‐R (scores), median (IQR)	4.0 (2.0, 7.3)	6.0 (3.0, 10.0)	Z = −2.267	0.023	5.0 (2.5, 9.0)	5.0 (2.0, 8.0)	Z = −0.442	0.678
TMT (scores), median (IQR)	223.5 (158.8, 331.2)	251.0 (182.0, 330.0)	Z = −0.931	0.361	219.0 (161.5, 294.5)	251.0 (184.5, 368.5)	Z = −1.405	0.111
DST (scores), median (IQR)	15.0 (14.0, 18.0)	12.0 (9.0, 15.0)	Z = −1.749	0.080	16.0 (14.0, 18.0)	16.0 (13.0, 19.0)	Z = −0.224	0.801
HVLT‐R_recall (scores), median (IQR)	3.0 (1.0, 4.0)	3.0 (1.0, 5.0)	Z = −0.217	0.829	3.0 (1.0, 5.0)	3.0 (1.0, 5.0)	Z = −0.590	0.591
HVLT‐R_recognition (scores), median (IQR)	22.0 (20.0, 23.0)	21.0 (19.0, 23.0)	Z = −0.926	0.355	22.0 (20.0, 23.0)	22.0 (19.0, 23.0)	Z = −0.490	0.625
BVMT‐R_recall (scores), median (IQR)	2.0 (1.0, 3.0)	2.0 (1.0, 4.0)	Z = −1.074	0.283	2.0 (1.0, 4.0)	2.0 (1.0, 3.5)	Z = −0.924	0.306
BVMT‐R_recognition (scores), median (IQR)	11.0 (9.8, 11.0)	11.0 (10.0, 12.0)	Z = −1.530	0.126	11.0 (10.0, 12.0)	10.0 (9.5, 12.0)	Z = −0.919	0.455
VFT (scores), median (IQR)	38.5 (32.8, 44.2)	43.0 (36.5, 52.5)	Z = −2.424	0.015	33.0 (28.0, 37.0)	34.0 (25.5, 39.0)	Z = −0.167	0.802

Note: The values in bold represent p values < 0.05 and the associated statistical values.

Abbreviations: ADL, activities of daily living; BMI, body mass index; BVMT‐R, Brief Visuospatial Memory Test–Revised; DST, Digit Span Test; HVLT‐R, Hopkins Verbal Learning Test–Revised; IQR, interquartile range; MMSE, Mini‐Mental State Examination; SD, standard deviation; TMT, Trail Making Test; VFT, verbal fluency test.

Primary outcome

Among the 131 eligible participants who completed the study, 35 (26.7%) developed dNCR within 1 week after surgery. The incidence of dNCR was 19.4% (12/62) in the olfactory enrichment group and 33.3% (23/69) in the sham group in the intention‐to‐treat analysis, with no statistically significant difference between groups (19.4% vs. 33.3%; χ ² = 3.259; p = 0.071; table 2). After adjusting for age, sex, years of education and baseline MMSE score, multivariable logistic regression confirmed a lower risk of dNCR in the olfactory enrichment group (incidence rate ratio 4.595, 95% confidence interval 1.428–14.788; β = 1.525; p = 0.011; table 2). No significant between‐group differences were observed in raw postoperative cognitive test scores (table S1).

TABLE 2.

Incidence rate ratio of dNCR between intervention and sham groups

	All eligible patients
	Overall (n = 131)	Intervention (n = 62)	Sham (n = 69)	Statistic	p value
dNCR incidence, n (%) a	35 (26.7)	12 (19.4)	23 (33.3)	χ 2 = 3.259	0.071

	Propensity score matched
	Overall (n = 82)	Intervention (n = 41)	Sham (n = 41)	Statistic	p value
dNCR incidence, n (%) a	21 (25.6)	5 (12.2)	16 (39.0)	χ 2 = 7.476	0.005
IRR (95% CI, unadjusted)	4.608 (1.494–14.213)	NA	NA	β coefficient = 1.528	0.008
IRR (95% CI, adjusted b )	4.595 (1.428–14.788)	NA	NA	β coefficient = 1.525	0.011

Note: The values in bold represent p values < 0.05 and the associated statistical values.

Abbreviations: CI, confidence interval; dNCR, delayed neurocognitive recovery; IRR, incidence rate ratio; NA, not applicable.

^a The differences between the sham group and the intervention group were compared by the chi‐square test.

^b Logistic regression models were adjusted for age, sex, years of education and baseline MMSE scores.

In the propensity‐matched cohort, the overall incidence of dNCR was 25.6% (21/82). The incidence was significantly lower in the olfactory enrichment group than in the sham group (12.2% vs. 39.0%; χ ² = 7.476; p = 0.005, table 2).

Secondary outcomes

Change scores for odour identification were calculated as postoperative minus preoperative scores. Participants with decreased odour identification scores (n = 32) had a higher incidence of dNCR than those with increased or unchanged odour identification scores (n = 99; 40.6% vs. 22.2%; χ ² = 4.183; p = 0.041; table 3). This association was not statistically significant in the PSM cohort.

TABLE 3.

dNCR incidence by olfactory function status

Olfactory function	Non‐dNCR, n (%)	dNCR, n (%)	Statistic	p value
All eligible patients
Decreased group (n = 32)	19 (59.4)	13 (40.6)	χ 2 = 4.183	0.041
Nondecreased group (n = 99)	77 (77.8)	22 (22.2)
Propensity score matched
Decreased group (n = 20)	14 (70.0)	6 (30.0)	χ 2 = 0.268	0.605
Nondecreased group (n = 62)	47 (75.8)	15 (24.2)

Note: The values in bold represent p values < 0.05 and the associated statistical values.

Abbreviation: dNCR, delayed neurocognitive recovery.

DISCUSSION

In this randomised trial of older adults undergoing total knee or hip replacement, perioperative olfactory enrichment reduced the incidence of dNCR, and postoperative olfactory decline was associated with a higher risk of dNCR. These findings provide preliminary clinical evidence supporting olfactory enrichment as a potential perioperative intervention and justify future randomised trials to further evaluate its efficacy. The prevalence of dNCR in our cohort was 26.7% in the full sample and 25.6% after propensity matching, consistent with prior reports. ⁵ , ³³ These findings align with preclinical evidence demonstrating that anaesthesia‐ and surgery‐induced olfactory impairment contributes to postoperative cognitive decline and that olfactory enrichment mitigates this effect in mice. ²¹ By translating this concept into a clinical setting, this study provides initial evidence supporting the therapeutic potential of olfactory enrichment for dNCR.

The beneficial effects of olfactory enrichment on dNCR are likely mediated through well‐characterised mechanisms of neural plasticity within the olfactory‐hippocampal network. Olfactory inputs travel through a direct, minimally synaptic pathway from the olfactory bulb to the hippocampus and entorhinal cortex, regions essential for memory formation and particularly vulnerable to perioperative stress. Evidence supporting this mechanistic link spans multiple domains. In rodents, olfactory enrichment reliably enhances neurogenesis in the dentate gyrus and promotes the survival of newborn neurones, changes associated with improved cognitive performance. ¹³ Enriched olfactory exposure also upregulates neurotrophic molecules such as brain‐derived neurotrophic factor, which enhance synaptic efficacy and support dendritic architecture in memory‐related circuits. ³⁷ Human neuroimaging studies further demonstrate that olfactory training enhances functional connectivity between olfactory regions and key components of the cognitive control and default mode networks. ³⁸ Novel odour stimulation additionally activates the locus coeruleus‐noradrenergic system, facilitating attention and memory consolidation. ³⁹ Given that anaesthesia and surgical trauma can transiently impair hippocampal function and disrupt synaptic homoeostasis, ⁴⁰ perioperative olfactory stimulation may act as a compensatory, resilience‐enhancing signal that helps preserve circuit integrity in the face of inflammatory and metabolic stress.

We also observed that participants with decreased odour identification scores were more likely to develop dNCR, suggesting that olfactory dysfunction may serve as both a risk marker and a mechanistic contributor to postoperative cognitive decline. Anaesthesia and surgery can induce inflammatory responses, including increased interleukin‐6 levels in blood and olfactory epithelium, which may impair olfactory receptor neurones. ²¹ Although it has been hypothesised that olfactory impairment may be associated with synaptic loss in the hippocampus, potentially contributing to cognitive impairment, ²¹ the causal link between these two phenomena has not been definitively established. It is also possible that anaesthesia and surgery independently induce both olfactory impairment and hippocampal synaptic loss without a direct causal relationship between them. Neuroimaging studies provide direct evidence for the neuroplastic effects of olfactory stimulation. Functional magnetic resonance imaging research demonstrates that olfactory training induces rapid reorganisation within olfactory‐cognitive networks. In a longitudinal study, Kollndorfer et al. found that 12 weeks of olfactory training increased activation in the piriform cortex and enhanced functional connectivity between the hippocampus and orbitofrontal cortex in patients with olfactory loss. ³⁸ This suggests strengthened communication between odour‐processing and memory‐related regions. Moreover, the mechanism likely involves odour‐driven hippocampal activation that promotes synaptic strengthening and neurotrophic support, potentially buffering against the neural stress of surgery and anaesthesia. Because inflammatory, tau, Aβ or acetylcholine markers were not measured, this sequence remains a plausible but unproven mechanism. Further research is needed to elucidate the precise mechanisms underlying these observations.

dNCR has been proposed to represent a postoperative neurodegenerative disease, ⁴¹ which is similar to AD with symptoms and risk factors. ⁴² Previous studies in patients with AD and olfactory dysfunction have reported reduced serum acetylcholine levels. This reduction impacts not just memory capabilities but is also crucial for the proper functioning of the olfactory system, suggesting that the diminished acetylcholine levels could be a contributing factor to the olfactory deficits observed in individuals with AD. ⁴³ Anaesthesia and surgery may also contribute to the decreased level of serum acetylcholine, ⁴⁴ which may, in turn, contribute to olfactory impairment in patients with dNCR. Future clinical studies are needed to investigate these potential neurochemical changes in perioperative patients.

Compared with studies of olfactory training in neurodegenerative diseases, our intervention differed substantially in timing, duration and target population. Prior work in individuals with mild cognitive impairment or AD typically involves long‐term interventions (12–24 weeks) and demonstrates modest benefits in select cognitive domains, including olfactory identification. In contrast, our perioperative intervention was brief and administered prophylactically. The significant reduction in dNCR observed after propensity matching indicates that the perioperative period, marked by heightened neuroinflammation and neural vulnerability, may represent a uniquely sensitive window during which olfactory enrichment can exert neuroprotective effects through mechanisms such as enhancing cognitive reserve or dampening neuroinflammatory cascades. Thus, our study advances a mechanistic hypothesis that targeting surgery‐induced olfactory dysfunction may help prevent downstream cognitive consequences.

We did not observe significant between‐group differences in raw cognitive scores or change scores across individual neuropsychological tests. Prior research using more intensive olfactory enrichment protocols (e.g., 40 odours twice daily for 15 days) has demonstrated improvements in attention, language and mood among older adults with dementia. ⁴⁵ The absence of comparable cognitive benefits in our study may be attributable to the short perioperative assessment window. Because cognitive testing was conducted only within 7 days after surgery, this timeframe may have been insufficient to detect the delayed or cumulative effects of olfactory enrichment on cognitive outcomes. It therefore remains to be determined whether longer periods of olfactory enrichment could further reduce the incidence of dNCR. Besides, the temporal relationship between olfactory decline and cognitive impairment remains unclear. Further studies are needed to explore the relationship between olfactory function and cognitive functions.

Our findings support the potential integration of olfactory enrichment into perioperative care as a simple, low‐cost and nonpharmacological strategy. For practical implementation, the intervention could follow the schedule used in our study, beginning 3 days before surgery and continuing through postoperative day seven, and could be incorporated into Enhanced Recovery After Surgery pathways without imposing additional clinical burden. Odour delivery kits containing mild, familiar essential oils (e.g., rose, lemon, eucalyptus) could facilitate standardised, twice‐daily passive exposure with minimal patient effort or staff involvement. Patients with mild preoperative olfactory impairment may particularly benefit from such interventions and could be prioritised based on brief preoperative screening. Given the low material costs and absence of specialised equipment, olfactory enrichment is highly scalable and adaptable to a wide range of healthcare settings, including resource‐limited environments. Future implementation research should evaluate not only cognitive outcomes but also feasibility metrics such as adherence and patient satisfaction, as well as clinical outcomes such as functional recovery and quality of life.

Limitation

This study has several limitations. First, olfactory function was assessed using a five‐odour identification array, which provides only a limited evaluation of olfactory ability and primarily reflects processes dependent on memory and language. ⁴⁶ Consequently, the observed association between reduced odour identification and dNCR should be interpreted cautiously. Future studies should incorporate comprehensive olfactory assessments—such as the Sniffin’ Sticks battery—which measure odour threshold, discrimination and identification and offer a more accurate characterisation of olfactory performance. Second, cognitive outcomes and dNCR incidence were assessed only within 1 week after surgery, restricting the ability to evaluate long‐term trajectories. Moreover, the study was conducted at a single centre and involved a specific surgical population, which may limit the generalisability of the findings. Further studies should include longer follow‐up periods (e.g., 6–12 months) and more diverse patient populations across multiple centres and surgical procedures.

Another limitation concerns blinding: Because olfactory enrichment is inherently sensory, participants may have inferred their group assignment, introducing potential expectancy or placebo effects. Although participants were informed only that they would be exposed to various inhaled substances, partial unblinding remains plausible, particularly in the absence of objective biomarkers or observer‐rated outcomes. Future studies should incorporate strategies to strengthen blinding or include objective physiological measures. Finally, the observed reduction in dNCR incidence was primarily evident in the propensity‐matched cohort and not in the full sample. Although the matched analysis provides supportive evidence for the potential effect of olfactory enrichment in mitigating dNCR, the findings should be interpreted with caution. Replications in larger, more diverse populations are needed to establish the generalisability and robustness of the results.

CONCLUSION

In conclusion, this study provides preliminary clinical evidence that perioperative olfactory enrichment may reduce the incidence of dNCR. The findings also suggest that anaesthesia‐ and surgery‐induced olfactory impairment may contribute to postoperative cognitive decline. Olfactory enrichment may represent a promising intervention for dNCR, potentially by mitigating perioperative olfactory dysfunction.

AUTHOR CONTRIBUTIONS

Xinchun Mei: Data curation; formal analysis; investigation; methodology; writing—original draft; funding acquisition. Zihan Ni: Data curation; writing—original draft. Shiyu Zhong: Data curation; writing—original draft. Jiayi Wang: Data curation; writing—original draft. Lin Zhu: Data curation; methodology; validation; writing—original draft. Zhongyong Shi: Data curation; methodology; validation; writing—original draft. Yupeng Chen: Investigation; writing—original draft. Hailin Zheng: Investigation; writing—original draft. Jingxiao Hu: Investigation; writing—original draft. Yuan Shen: Conceptualization; funding acquisition; methodology; supervision; writing—review and editing.

FUNDING

This study was supported by the National Natural Science Foundation of China (82320108005; Yuan Shen), the Science and Technology Commission of Shanghai Municipality (23XD1403200; Yuan Shen), the Fundamental Research Funds for the Central Universities (YG2024LC11; Yuan Shen) and Shanghai Mental Health Center (2025‐QM04; Xinchun Mei).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENTS

Ethical approval for this study was obtained from the Clinical Research Ethics Committee of the Shanghai 10th People's Hospital, Shanghai, China (approval no. SHSY‐IEC‐2.0/15–78/01).

TRIAL REGISTRATION

This trial was registered at ClinicalTrials.gov (NCT03441074) on 15 February 2018. Data collection was completed in December 2019. The 2025 update in ClinicalTrials.gov reflects a post hoc addition of the statistical analysis plan to the registry record; no participant‐related procedures were added or modified after 2019.

Supporting information

Supporting Information S1

Table S1

Biography

Xinchun Mei obtained her Doctor of Medicine degree in psychiatry and mental health from Tongji University School of Medicine, China, in 2020. She is currently an attending psychiatrist at the Shanghai Mental Health Center in China. Her main research interests include biomarkers of perioperative neurocognitive disorders and related nonpharmacological therapies.

DATA AVAILABILITY STATEMENT

Participant data and statistical analysis code will be made available upon reasonable request to the corresponding author, after the approval of a proposal and a signed data access agreement. Data will be shared in accordance with institutional and ethical guidelines.